Filling the crucial gaps inside the AI Infrastructure ecosystem, to make it commercially viable.

devto

productHunt

substack

medium

github

Palash Bagchi  • Peerlist

<h3>Kakunin provides the accountability layer for AI agents, enabling them to operate autonomously and securely within regulated industries. It issues cryptographic identities (X.509 certificates) via AWS KMS, allowing AI agents to have a verifiable identity, similar to how financial institutions use X.509 for secure transactions.</h3><ul><li>Cryptographic Identity: Issue <a target="_blank" rel="noopener noreferrer" class="break-all link-break text-gray-500 hover:underline" href="https://www.kakunin.ai/">X.509 certificates for AI agents</a>, binding their identity, operator, and permitted actions. Private keys are securely stored in AWS KMS.</li><li>Real-time Behavioural Monitoring: Continuously monitor agent actions and outputs for drift and anomalies. Detect deviations from baseline behavior and flag them with a rolling 30-day risk score.</li><li>Content-Risk Scoring: Evaluate the content of agent outputs to flag harmful, prohibited, or off-scope language, adhering to regulations like the EU AI Act.</li><li>Automated Revocation: When an agent's risk score crosses a configurable threshold (default 0.85), its certificate is automatically revoked within 60 seconds, preventing further unauthorized actions.</li><li>Compliance Reporting: Generate regulator-ready compliance reports mapped to frameworks such as NIST AI RMF, NIST CSF 2.0, ISO 27001, MiCA, and the EU AI Act. Reports are available in PDF and JSON formats.</li><li>Audit Trails: Maintain an immutable, append-only audit log of all agent actions, decisions, and events, ensuring a complete chain of custody for regulatory scrutiny.</li><li>Integration: Easily integrate Kakunin into your systems via a comprehensive REST API and a fully typed TypeScript SDK. Features include webhook event delivery with HMAC signatures and a sandbox mode for testing.</li><li>AgentMail: Each certified agent receives a verifiable email address, facilitating communication with regulators and counterparties, with all messages logged in the audit trail.</li><li>Public Verification: A public endpoint allows anyone to verify an agent's identity, scope, and revocation status in under 500ms without authentication.</li><li>Developer Tools: Benefit from features like Zod-validated responses, automatic retries, webhook signature verification, and OTLP export to popular observability platforms.</li></ul>Kakunin is designed for financial institutions, fintechs, and any organization deploying AI agents in regulated environments, ensuring they can operate with autonomy while maintaining auditable control and compliance.